It'll be funny if it turns out that the neurons are actually quite smart and their only problem is they don't have much in the way of arms, legs etc to prove it:).
They claim they're giving you options to work profitably with people in "LAGOS, NIGERIA", or alternative ways of spending your money;).
Anyway, my point is too often people think that having options is always good, and that's FALSE. Many people have been brainwashed by their society/media to think that way, but it's a fallacy.
Having two options instead of one is NOT always good.
Giving a person one good option and one bad option can often be worse than giving them one good option (or even making the decision for them without bothering them with it).
And if they have to make the decision anyway, it matters little whether it's spammed into their inbox, or they end up in the store, trying to decide what to buy.
I KNOW it's not usually "black and white" like that, I'm just pointing out it's wrong to think that options/alternatives are automatically good.
An alternative to Apple's store might be good. Or it might not be - depends on the store, depends on Apple's response, depends on what happens.
> It's generally good for society to be presented alternatives.
Only if the alternatives aren't crap. Better useful options = good. Only one option = usually bad (but not always;) ). Lots of crappy options = bad.
Having lots of spam in mailboxes providing people with more choices = bad.
More crappy options = higher chance for people to make the wrong decisions. If a user interface presented new users with zillions of choices, choices that they know nothing about it does not help them. Pushing alternatives to a screen that "noobs" don't see (and will "never" encounter on their own) = making the choice for them = less choices for them.
GC makes it easier. People can still screw up, but you are doing away with an entire class of too common mistakes.
Using C is just like driving a manual without a clutch. It can be done, you just need to concentrate. One mistake and poof. So far there appear to be only a handful of people in the entire world who can write safe C. So far Dan Bernstein appears to write reasonably safe C and even he makes some mistakes ( http://www.jcb-sc.com/qmail/guninski.html )...
Using Java/C# is like driving an automatic. And advanced "JIT" VMs are like those new high performance automatic transmissions - they can do better than most humans AND more importantly they are more consistent and reliable when doing it.
FWIW, I wonder why it is common to push data onto an address stack. To me address = code. Mixing data and code ( addresses ) is bad hygiene. You should only do it in controlled circumstances, not as a norm. Why can't you have separate stacks? While there will still be problems, it will make certain bugs harder to exploit.
Say you are using water power to spin something at a certain speed.
Voltage is like water pressure.
Amps is the amount of water flowing per second.
To transmit the same amount of power, you could have a thick pipe, low pressure and lot of water flowing. Or a skinny pipe, very high pressure and very little water flow (at the other end water at high pressure would be able to spin a turbine at the same speed as the low pressure water with lots of water despite there being a lot less water).
From the article: "Facebook's new distribution scheme calls for 277 volt power to the servers. "We're working with power supply vendors to create a (server) power supply that will accept 277 volts on the input," said Michael."
"Webserver(s) and single DB" is a common configuration because it is convenient. Not because of slow languages. In fact sites often start with just one webserver and one DB and never ever notice the "slow language" problem. This configuration typically allows people to easily add webservers if they have to, but doesn't allow them to easily add DB servers.
The convenient DBs are actually quite slow and don't scale well (as the AC replying to you has noticed;) ).
The scalable DBs that the huge sites use are typically not easy to use or very expensive (or both) or lack a lot of features (so you have to reproduce the features you want in the application).
Furthermore, to do what ebay, amazon and facebook do requires a lot of "roll your own" and customization [1]. Most of the huge sites "shard" almost everything and aim for "shared nothing" architectures, and then have lots of custom "glue and tape" to hold stuff together.
So the fact that facebook is experiencing problems with PHP doesn't really prove much (other than incompetence). Stuff regularly breaks on facebook, they're probably doing lots of things wrong (and just enough things right;) ). Using PHP is one of the things they are doing wrong, but for it to be such a huge problem seems very weird to me. Since if you design stuff right, you should be able to rewrite the slow CPU-sucking bits in a faster language (or gradually change everything to a different language).
[1] None of them seem to have the same system. In contrast thousands of tiny sites are happily running diverse apps on "LAMP".
The thing is, many webapps are actually DB bound despite appearing to be webserver bound, because they store most of their state in the database (and nearly the rest of it at the user's browser).
Now this means you are shifting the burden of locking, serialization and other tricky stuff to the DB.
When you do that, you can have as many identical webservers as you want (scaling "horizontally")- since the state is all at the DB (and most of the rest in the user's browser).
You can then use a crappy slow language for your webapp, since it doesn't matter, even if it does look like your webserver becomes cpu bound, you can easily add webservers if you need, to increase capacity.
But it is not as easy to add DB servers to speed stuff up or increase capacity.
Because most locking and serialization stuff doesn't scale horizontally very well. A node has to either wait for the rest of the nodes to say a write is OK (which is slow), or some funky stuff has to be done to resolve conflicts, or stuff has to be sharded (conflict avoidance - nodes are never responsible for the same thing, so there are no conflicts). That's why those cluster supercomputers use expensive fast low latency interconnects. And that's why databases always have compromises when you want to do multimaster clusters.
So you try to shift as much cpu usage (stateless checking, calculations, caching etc) to the webapps and just leave the locking, serialization and transaction nastiness to the DB.
And that's why despite it being nice and consistent to run stored procedures in the database to handle lots of stuff, it becomes prohibitive/expensive in practice to do so when you have such architectures.
Nah, the selling point is the low latency, "no tearing" scrolling on the iphone that just follows your finger (and my guess is psychologically more likely to be treated as part of your body[1]);).
If you look at the scrolling on other phones you'll find they lag a bit, or tear a bit, or are a bit jerky. That breaks the illusion/magic/coolness.
I can imagine Steve Jobs (or a lesser incarnation) yelling the developers if they try to get away with the jerky stuff you see on other phones...
It's stuff like that that makes many Apple products different.
They're not the most functional, or most reliable (and definitely not more secure). But some parts are very very polished.
[1] You don't easily give up parts of your body even if they are flawed in some ways.
Unfortunately that's not always the case - the person who writes the software might find its usage obvious and not know how to explain it well to others.
And often the users who find it unobvious, and finally figure out by themselves how to use something, forget which parts were unobvious. They weren't taking down notes on the unobvious bits - typically too busy just trying to get things to work. So it might be useful to have some people taking notes on "questions" they have that aren't answered by the documentation, and other problems, while learning the stuff.
Sometimes in OSS the developers appear to have the opinion that the source code is good enough documentation. I've also seen the ridiculous case in Linux of a well documented method/API being deprecated in favour of a poorly documented new API/method (which may be technically better, but only a few in the world can figure out how to use it).
In general things in FreeBSD tend to be better documented than in a typical Linux distro. At least that's what I find from comparing the manpages (yes I've tried using the "info" docs on suse and ubuntu, and too often I get more useful help from "experts exchange";) ).
Oh yah, in case it wasn't obvious, virtual telekinesis is also part of the human augmentation I'm talking about.
Thinking of something could cause your wearable computer to perform an action. Which could include visiting a "favorite" url (which could result in turning the lights on or off, or setting the temperature of the airconditioning etc). Or running a script/program.
FWIW, I'm not sure how many would bother to take the trouble to train their wearable (and their minds[1]) to "virtually type" stuff quickly. If most don't, keyboards may still remain popular (and some people might have hard to distinguish thought patterns).
[1] They need to think up many unique thought macros and be able to remember them and "rethink" them quickly in order to type and _edit_ fast.
For quick "typing" you'd need a macro for each common word. It may be just the thought of the word itself, so not as tedious as thinking up new outlandish stuff - but you still need to spend time associating them.
Then you could use "thought macros" to control wearable computers.
The measurements of thought patterns are likely to be specific to each person. So devices that use thought input would have to be trained.
But after that, you could be thinking of stuff like "purple green striped elephant" as the escape sequence to tell the computer to start listening in and doing stuff based on the thought patterns it recognizes (which could include: take a picture of this, and associate it with this current thought pattern). You then use a different unique thought pattern to get it to stop (or press a manual button if stuff screws up:) ).
Then whenever you think of a thought pattern the picture or other computer memorized object (audio, url, file) will be recalled.
I have long considered this as the next step in augmenting humans.
Once you have that, people will have perfect photographic, "videographic" and "audiographic" memory, and be capable of "virtual telepathy" e.g. communicate with others just by thinking (which is trivial obvious step once you combine the thought macro stuff with wireless comms).
The problem people have to realize is the MPAA, RIAA, DRM and restrictive copyright laws will get in the way of such augmentation. It will either be crippled, prohibited or taxed severely.
They may not be happy with just a "penny for your thoughts", especially if they have got the law to consider it as "their thoughts" and their property, and not yours.
p.s. I don't see this development as "innovative", since I consider it a rather obvious step (along with the other rather obvious steps I've mentioned in this post), but I'm sure it's probably patented etc;).
> Passing through this little valley took so long, climbing up that hill the rhythm changed
Which brings us to the question: Are the ants really counting steps or is it based on timing?
Because you could get a similar result if it's based on timing. For example the ant walks to point X and it takes 60 seconds, so on the return journey, somewhere in the ants brain there's a countdown from 60 seconds (or more likely a increase/decrease in "potential"). With longer, but not much heavier legs the ant could still overshoot the nest because it's moving faster.
So what if you increased the slope so that the ant slowed down (or sped up) while taking the same number of steps for a particular distance. Would that change where the ant started looking for the nest? The difficulty is the ant could compensate for that - if something takes longer the "meter/timer" could be paused.
Also there's a more "analog" way of counting and there's a more "digital" way of counting.
Digital is the hard edged "1, 2, 3" count. Analog would be closer to filling a pool with buckets of water when counting up and emptying it when counting down.
If they run a cable to cross a river/chasm via a bridge, it's going to cost a bit to build another bridge just for your cable. They could take the "down" route, but that's not cheap either...
Satellite = high latency. Wireless could be an option (but usually still involves cables at some point).
Yeah, and the problem might be overblown (and the reason why it is overblown could be politics or part of "haggling").
100 hours in five weeks might be not much depending on how many branches there are. Same goes for the 4,677 hours in 6 months figure.
The branches may just have ADSL links to the HQ.
As for redundant network links. In many cases it's worthless paying the extra. It costs a lot to do it right.
1) The branch may only have one telco/carrier choice for connectivity (at a reasonable price) 2) Even if they have two choices, both could go through the same line/chunk of ground that the backhoe digs up. Nowadays with all the outsourcing and inter company deals, you might buy redundant links from two different companies, and later find out the hard way that they are all in the same cable!
Many years ago my ISP had two redundant international links (they were small then). But they lost connectivity even, so people asked "what happened". And they said - a trawler broke one line, and apparently not long later, another trawler (or the same trawler:) ) broke the second line[1]... Doh.
And I wonder how many paid for redundant transpacific fibre optic links that all went over the Taiwan earthquake zone;).
[1] That's what they said anyway... For all we know, maybe it was the NSA doing a bad job of tapping the lines with their subs:).
Slashdot Mirror
I suspect my uncle's blood type is XO positive.
Here's how you improve a process at a hospital:
http://www.post-gazette.com/pg/06318/738252-114.stm
Yeah. Maybe your example should be hundreds of Chinese, Indonesian etc merchant ships in San Francisco Bay loaded up with weapons.
Make it more obvious.
It'll be funny if it turns out that the neurons are actually quite smart and their only problem is they don't have much in the way of arms, legs etc to prove it :).
They claim they're giving you options to work profitably with people in "LAGOS, NIGERIA", or alternative ways of spending your money ;).
Anyway, my point is too often people think that having options is always good, and that's FALSE. Many people have been brainwashed by their society/media to think that way, but it's a fallacy.
Having two options instead of one is NOT always good.
Giving a person one good option and one bad option can often be worse than giving them one good option (or even making the decision for them without bothering them with it).
And if they have to make the decision anyway, it matters little whether it's spammed into their inbox, or they end up in the store, trying to decide what to buy.
I KNOW it's not usually "black and white" like that, I'm just pointing out it's wrong to think that options/alternatives are automatically good.
An alternative to Apple's store might be good. Or it might not be - depends on the store, depends on Apple's response, depends on what happens.
> It's generally good for society to be presented alternatives.
Only if the alternatives aren't crap. Better useful options = good. Only one option = usually bad (but not always ;) ). Lots of crappy options = bad.
Having lots of spam in mailboxes providing people with more choices = bad.
More crappy options = higher chance for people to make the wrong decisions. If a user interface presented new users with zillions of choices, choices that they know nothing about it does not help them. Pushing alternatives to a screen that "noobs" don't see (and will "never" encounter on their own) = making the choice for them = less choices for them.
And if happiness is considered an important factor see:
http://www.youtube.com/watch?v=LTO_dZUvbJA
http://www.youtube.com/watch?v=VO6XEQIsCoM
And: http://en.wikipedia.org/wiki/The_Paradox_of_Choice
GC makes it easier. People can still screw up, but you are doing away with an entire class of too common mistakes.
Using C is just like driving a manual without a clutch. It can be done, you just need to concentrate. One mistake and poof. So far there appear to be only a handful of people in the entire world who can write safe C. So far Dan Bernstein appears to write reasonably safe C and even he makes some mistakes ( http://www.jcb-sc.com/qmail/guninski.html )...
Using Java/C# is like driving an automatic. And advanced "JIT" VMs are like those new high performance automatic transmissions - they can do better than most humans AND more importantly they are more consistent and reliable when doing it.
FWIW, I wonder why it is common to push data onto an address stack. To me address = code. Mixing data and code ( addresses ) is bad hygiene. You should only do it in controlled circumstances, not as a norm. Why can't you have separate stacks? While there will still be problems, it will make certain bugs harder to exploit.
Yes that's kind of how it sounds depending on the water pressure.
Say you are using water power to spin something at a certain speed.
Voltage is like water pressure.
Amps is the amount of water flowing per second.
To transmit the same amount of power, you could have a thick pipe, low pressure and lot of water flowing. Or a skinny pipe, very high pressure and very little water flow (at the other end water at high pressure would be able to spin a turbine at the same speed as the low pressure water with lots of water despite there being a lot less water).
From the article: "Facebook's new distribution scheme calls for 277 volt power to the servers. "We're working with power supply vendors to create a (server) power supply that will accept 277 volts on the input," said Michael."
Why 277 volts?
The ones I went to didn't use newspapers either ;).
"Webserver(s) and single DB" is a common configuration because it is convenient. Not because of slow languages. In fact sites often start with just one webserver and one DB and never ever notice the "slow language" problem. This configuration typically allows people to easily add webservers if they have to, but doesn't allow them to easily add DB servers.
;) ).
;) ). Using PHP is one of the things they are doing wrong, but for it to be such a huge problem seems very weird to me. Since if you design stuff right, you should be able to rewrite the slow CPU-sucking bits in a faster language (or gradually change everything to a different language).
The convenient DBs are actually quite slow and don't scale well (as the AC replying to you has noticed
The scalable DBs that the huge sites use are typically not easy to use or very expensive (or both) or lack a lot of features (so you have to reproduce the features you want in the application).
Furthermore, to do what ebay, amazon and facebook do requires a lot of "roll your own" and customization [1]. Most of the huge sites "shard" almost everything and aim for "shared nothing" architectures, and then have lots of custom "glue and tape" to hold stuff together.
So the fact that facebook is experiencing problems with PHP doesn't really prove much (other than incompetence). Stuff regularly breaks on facebook, they're probably doing lots of things wrong (and just enough things right
[1] None of them seem to have the same system. In contrast thousands of tiny sites are happily running diverse apps on "LAMP".
The thing is, many webapps are actually DB bound despite appearing to be webserver bound, because they store most of their state in the database (and nearly the rest of it at the user's browser).
Now this means you are shifting the burden of locking, serialization and other tricky stuff to the DB.
When you do that, you can have as many identical webservers as you want (scaling "horizontally")- since the state is all at the DB (and most of the rest in the user's browser).
You can then use a crappy slow language for your webapp, since it doesn't matter, even if it does look like your webserver becomes cpu bound, you can easily add webservers if you need, to increase capacity.
But it is not as easy to add DB servers to speed stuff up or increase capacity.
Because most locking and serialization stuff doesn't scale horizontally very well. A node has to either wait for the rest of the nodes to say a write is OK (which is slow), or some funky stuff has to be done to resolve conflicts, or stuff has to be sharded (conflict avoidance - nodes are never responsible for the same thing, so there are no conflicts). That's why those cluster supercomputers use expensive fast low latency interconnects. And that's why databases always have compromises when you want to do multimaster clusters.
So you try to shift as much cpu usage (stateless checking, calculations, caching etc) to the webapps and just leave the locking, serialization and transaction nastiness to the DB.
And that's why despite it being nice and consistent to run stored procedures in the database to handle lots of stuff, it becomes prohibitive/expensive in practice to do so when you have such architectures.
Nah, the selling point is the low latency, "no tearing" scrolling on the iphone that just follows your finger (and my guess is psychologically more likely to be treated as part of your body[1]) ;).
If you look at the scrolling on other phones you'll find they lag a bit, or tear a bit, or are a bit jerky. That breaks the illusion/magic/coolness.
I can imagine Steve Jobs (or a lesser incarnation) yelling the developers if they try to get away with the jerky stuff you see on other phones...
It's stuff like that that makes many Apple products different.
They're not the most functional, or most reliable (and definitely not more secure). But some parts are very very polished.
[1] You don't easily give up parts of your body even if they are flawed in some ways.
Unfortunately that's not always the case - the person who writes the software might find its usage obvious and not know how to explain it well to others.
;) ).
And often the users who find it unobvious, and finally figure out by themselves how to use something, forget which parts were unobvious. They weren't taking down notes on the unobvious bits - typically too busy just trying to get things to work. So it might be useful to have some people taking notes on "questions" they have that aren't answered by the documentation, and other problems, while learning the stuff.
Sometimes in OSS the developers appear to have the opinion that the source code is good enough documentation. I've also seen the ridiculous case in Linux of a well documented method/API being deprecated in favour of a poorly documented new API/method (which may be technically better, but only a few in the world can figure out how to use it).
In general things in FreeBSD tend to be better documented than in a typical Linux distro. At least that's what I find from comparing the manpages (yes I've tried using the "info" docs on suse and ubuntu, and too often I get more useful help from "experts exchange"
Yep. The tech is getting there. Very slowly though IMO.
But in the near future I think the legal crap will be the thing that is slowing progress.
winkey+pause is good enough to determine version for most windows support issues.
/all is important ;).
Too bad in many support calls, running ipconfig
The number of users who still would need to run "winipcfg" are dwindling... Too bad microsoft didn't leave it in for later versions of windows.
> The variance in electrical activity in the brain between two people receiving the same sensory input is, in an average way, too great to be useful.
You shouldn't directly use the "thought pattern" of a person to tag the data.
As you said, the thought patterns are likely to be different from person to person.
What you do though is, for each tagging participant, you get the person's thought patterns for a whole bunch of tags.
Then they can tag stuff really quickly just by looking at them. The advanced people might be able to manage multi-tags with a single thought pattern.
Lastly, I don't think you necessarily have to use a thought pattern that's related to the object.
Oh yah, in case it wasn't obvious, virtual telekinesis is also part of the human augmentation I'm talking about.
Thinking of something could cause your wearable computer to perform an action. Which could include visiting a "favorite" url (which could result in turning the lights on or off, or setting the temperature of the airconditioning etc). Or running a script/program.
FWIW, I'm not sure how many would bother to take the trouble to train their wearable (and their minds[1]) to "virtually type" stuff quickly. If most don't, keyboards may still remain popular (and some people might have hard to distinguish thought patterns).
[1] They need to think up many unique thought macros and be able to remember them and "rethink" them quickly in order to type and _edit_ fast.
For quick "typing" you'd need a macro for each common word. It may be just the thought of the word itself, so not as tedious as thinking up new outlandish stuff - but you still need to spend time associating them.
It's not going to replace keyboard tagging now.
:) ).
;).
But in the future more advanced versions might.
Then you could use "thought macros" to control wearable computers.
The measurements of thought patterns are likely to be specific to each person. So devices that use thought input would have to be trained.
But after that, you could be thinking of stuff like "purple green striped elephant" as the escape sequence to tell the computer to start listening in and doing stuff based on the thought patterns it recognizes (which could include: take a picture of this, and associate it with this current thought pattern). You then use a different unique thought pattern to get it to stop (or press a manual button if stuff screws up
Then whenever you think of a thought pattern the picture or other computer memorized object (audio, url, file) will be recalled.
I have long considered this as the next step in augmenting humans.
Once you have that, people will have perfect photographic, "videographic" and "audiographic" memory, and be capable of "virtual telepathy" e.g. communicate with others just by thinking (which is trivial obvious step once you combine the thought macro stuff with wireless comms).
The problem people have to realize is the MPAA, RIAA, DRM and restrictive copyright laws will get in the way of such augmentation. It will either be crippled, prohibited or taxed severely.
They may not be happy with just a "penny for your thoughts", especially if they have got the law to consider it as "their thoughts" and their property, and not yours.
p.s. I don't see this development as "innovative", since I consider it a rather obvious step (along with the other rather obvious steps I've mentioned in this post), but I'm sure it's probably patented etc
> Passing through this little valley took so long, climbing up that hill the rhythm changed
Which brings us to the question: Are the ants really counting steps or is it based on timing?
Because you could get a similar result if it's based on timing. For example the ant walks to point X and it takes 60 seconds, so on the return journey, somewhere in the ants brain there's a countdown from 60 seconds (or more likely a increase/decrease in "potential"). With longer, but not much heavier legs the ant could still overshoot the nest because it's moving faster.
So what if you increased the slope so that the ant slowed down (or sped up) while taking the same number of steps for a particular distance. Would that change where the ant started looking for the nest? The difficulty is the ant could compensate for that - if something takes longer the "meter/timer" could be paused.
Also there's a more "analog" way of counting and there's a more "digital" way of counting.
Digital is the hard edged "1, 2, 3" count. Analog would be closer to filling a pool with buckets of water when counting up and emptying it when counting down.
No, they just lob them.
Soon after you get arrested for a crime you didn't commit? ;)
I didn't go into the details in my original post but here's some evidence to back up what I said about network redundancy:
http://www.washingtonpost.com/ac2/wp-dyn/A23689-2003Jul7?language=printer
http://www.wired.com/science/discoveries/news/2006/01/70040?currentPage=all
If they run a cable to cross a river/chasm via a bridge, it's going to cost a bit to build another bridge just for your cable. They could take the "down" route, but that's not cheap either...
Satellite = high latency. Wireless could be an option (but usually still involves cables at some point).
Even the spooks have troubles with diggers: http://www.washingtonpost.com/wp-dyn/content/article/2009/05/30/AR2009053002114_pf.html
But they sure have a bunch of alert people monitoring their lines (which are probably used to monitor your lines ;) ).
Yeah, and the problem might be overblown (and the reason why it is overblown could be politics or part of "haggling").
:) ) broke the second line[1]... Doh.
;).
:).
100 hours in five weeks might be not much depending on how many branches there are. Same goes for the 4,677 hours in 6 months figure.
The branches may just have ADSL links to the HQ.
As for redundant network links. In many cases it's worthless paying the extra. It costs a lot to do it right.
1) The branch may only have one telco/carrier choice for connectivity (at a reasonable price)
2) Even if they have two choices, both could go through the same line/chunk of ground that the backhoe digs up. Nowadays with all the outsourcing and inter company deals, you might buy redundant links from two different companies, and later find out the hard way that they are all in the same cable!
Many years ago my ISP had two redundant international links (they were small then). But they lost connectivity even, so people asked "what happened". And they said - a trawler broke one line, and apparently not long later, another trawler (or the same trawler
And I wonder how many paid for redundant transpacific fibre optic links that all went over the Taiwan earthquake zone
[1] That's what they said anyway... For all we know, maybe it was the NSA doing a bad job of tapping the lines with their subs